Email is a huge space. So many things are broken today that it's hard to know where to even start fixing. I'm very proud of what we've achieved at Xobni but there are many exciting things beyond what our sidebar does - plenty of problems to attack.
The email experience can be improved so much with just simple data mining like collecting information about a project by just looking at project related correspondence. Who has been active or who is not. Great deal of information about user is hidden in those mails, I am sure an intelligent pass through them can tell simple things like gender of the person, career, location, purchase history, relationship status and much more.
Also, another thing makes me wonder how email clients will adopt themselves to manage messages in forms other than text.
They had a lot of great ideas and vision but the project didn't turn out to be a success unfortunately.
I think there is a book called 'Dreaming in Code' which documents Chandler's failure in more detail. Probably worth a read! :)
They had this notion of items - everything is an item and you can stamp an item into an item of a different kind. For instance, you get stamp a mail as a TODO item.
Some of my facebook messages, some twitter messages and some RSS feeds (but not all of them) are more important than some (but not all) of my emails. Yet, the facebook/twitter/email divergence is also a part of my consciousness as a user. To identify these similar things across mediums, present these things in once place and yet subtly distinguish between them - this is the challenge.
I guess it is also easy for me to point out that the answer is a combination of machine learning/statistical analysis and good UI design. Implementing it is the hard part :) I have thought about it but I am a bit busy doing other things. It would be interesting to hear your take on this and your approach to this problem when you are able to reveal such things.